Video editing... X5677 vs X5690

[tpivette89]

OK, before I begin my question, let me say this: I did an extensive search with these two CPUs in mind, and couldn't find a definitive answer on the net or these forums.

I have an early 2009 Mac Pro (DP config, flashed to 5,1 running Mojave with a HD 7970). Looking to upgrade the processors and was wondering if I could gain anything choosing a pair of X5690s over a pair of X5677s. For comparison, the X5677s are about $38 a pair vs a pair of X5690s run around $180.

I plan to use the computer to run either iMovie or FCPX... would the extra 4 cores for $142 be worth it (non-professional) vs the 8 cores running the same clock speed?

Thanks SO MUCH in advance. I have the computer ready to tear down based on input received...

PS: I have delidded many CPUs, so this process doesn't bother me. I'm just simply looking for the advantages between the 8 core vs 12 core if the clock speeds are the same vs the definitive increase in cost of the pair of the CPUs?

 

[bsbeamer]

what are you on now?
what other upgrades have you made to the system? SSDs, RAM, etc?

Personally have 2 x 3.46 on 5,1 with 128GB RAM. In Adobe applications all day long, mostly on video. Happy I upgraded to this for AE purposes. Not sure it mattered as much for PPro over a slightly slower processor.

 

[flowrider]

^^^^That wasn't his question. Both CPUs the OP is talking about are 3.46MHz. The X5677 has 4 cores and the X5690 has 6 cores. So with a DP machine, either 8 or 12 cores. Personally, I have the X5677s, paid a lot more for them when I bought them For what I do, 8 cores are plenty.

But, yes, IMO, an SSD or two or three, preferably running in the PCI slots are also necessary.

Lou

 

[crjackson2134]

I plan to use the computer to run either iMovie or FCPX...

I’m not an expert here, but yes, it’s probably worth the difference if you do more than an occasional short edit/render in these applications. Perhaps I should say, it would be worth it to me. Just my opinion...

 

[mikas]

From Your signature:
"2009 Mac Pro: 2 x W3520 (6 core, 2.26ghz), 24gb 1333mhz RAM, Radeon HD 7970"
Shoul it be a E5520 instead? W3520 is a single core only Xeon (QPI links: 1).

I upgraded from 2x E5620 to 2x X5680. Used X5680 is a lot cheaper than X5690, and the speed difference is more like nominal. I am pleased with my results (see screenshots).

I don't do much video, but mostly 3D renders. Cinerender takes all the cores it can get and performs accordingly. If your software can fully take advantage of all processor cores, then I would recommend the upgrade with 6-core models. If your software is not properly multithreaded, it's a little bit different story.

If your processors are E5520 as I suggested, you would get even more of an upgrade than I did. That is if your software can take advantage of all of the cores.

Then there are the graphics cards to accelerate your work too. But that's for someone else to take on that.

 

[SneakyTLoD]

IDK, should this really be a question anymore? Get the fastest possible at this point as you'll never have to do it again. We're end of life here and $150 difference can't be that much to you right? The X5690's are the top of the line with the 4,1&5,1. The amount of time it takes to upgrade 2 times has to be worth more than the $150. Just max out or wait until the next model at this point. If you are only doing single thread work I'd bet you'd have moved on at this point but if not the X5687 would be a tad faster. I doubt you'd notice a difference in single core work.

BTW de-lidding Xeon's is a bit different than de-lidding standard Intel "i" processors as the Xeons are soldered. Just FYI.

I noticed more of an improvement in Video and Photo work with the addition of an NVMe drive than the processors and ram. With audio it was quite the other way. Just depends on your needs. I still say max it out at this point as we're in cheap mode. Once the new ones are out we're looking at $8k for a well equipped new machine. If you are cheap like me just max out your 5,1 and wait for a used 7,1 one to become affordable.

 

[DPUser]

TL;DR:

The real question is "How well do your applications utilize multiple cores?" I run a recording studio. My DAW of choice uses all 12 cores of my Mac Pro efficiently.

As for X5690 vs X5680, if paying twice as much to gain 3.6% in speed is worth it to you, go for it.

 

[orph]

no
"I plan to use the computer to run either iMovie or FCPX... would the extra 4 cores for $142 be worth it (non-professional) vs the 8 cores running the same clock speed?"

XD simple answer

if you have the spare cash you can, it will at times be faster

if it was me on a budget id get the X5677

and if it was me and i was on a budget i might look at the dual 4 core and then put the money towards SSD/hard drives or something (camera kit, lens, tripod ect)

but if you have the cash and want the fastest you can get go for it,

 

[handheldgames]

OK, before I begin my question, let me say this: I did an extensive search with these two CPUs in mind, and couldn't find a definitive answer on the net or these forums.

I have an early 2009 Mac Pro (DP config, flashed to 5,1 running Mojave with a HD 7970). Looking to upgrade the processors and was wondering if I could gain anything choosing a pair of X5690s over a pair of X5677s. For comparison, the X5677s are about $38 a pair vs a pair of X5690s run around $180.

I plan to use the computer to run either iMovie or FCPX... would the extra 4 cores for $142 be worth it (non-professional) vs the 8 cores running the same clock speed?

Thanks SO MUCH in advance. I have the computer ready to tear down based on input received...

PS: I have delidded many CPUs, so this process doesn't bother me. I'm just simply looking for the advantages between the 8 core vs 12 core if the clock speeds are the same vs the definitive increase in cost of the pair of the CPUs?

Dual x5680 is the way to go. Rendering video transitions, outputting projects, editing, etc. scales well across multiple CPUs. Having 4 additional cores than the x5677 at a slightly slower cpu speed, the x5680 provides additional bandwidth that the x5677 can't muster.

This may give you an idea as to what to expect from a dual x5677(geekbech) and a dual x5680(geekbench)

The dual x5680 is worth the additional $50.

 

[pertusis1]

I'm not sure. I went straight from 2.4 *12 to 3.46 *12. That was a drastic difference. FWIW, FCPX does use all of the available cores with some hyperthreading as well, though it doesn't often run them at 100%.

 

[tpivette89]

Thanks for all the replies!

Expanding on my setup info... I have a Evo 860 (250GB) partitioned with High Sierra on one and Mojave on the other, and another 860 EVO (1TB) as my media storage drive. I plan on adding a PCIe 970 EVO as a video editing scratch disc. Currently I have only 24GB of RAM, but plan on doubling that before I finish the upgrades.

To be perfectly honest, I already have a pair of X5690s that I delidded and installed, but have not done any video editing with them yet. After seeing what delidded ones go for on Ebay, I began to wonder if I needed all 12 cores to suit my needs. I could sell those, and fund the rest of my upgrades. At the crazy low price the X5677s are going for, I ordered a set to try out. I will eventually try editing with both sets of processors and will determine from there if the speed bump from 4 extra cores is worth what I could sell the delidded X5690s for.

This may give you an idea as to what to expect from a dual x5677(geekbech) and a dual x5680(geekbench)

Honestly, I didn't think a pair of X5677s would score 20000... let alone 21000. Seeing that score confirmed my resolve to purchase a pair and at least test them out. Thanks for that link.

 

[flowrider]

^^^^Just for $hits and giggles I just ran mine:

Lou

 

[tpivette89]

^^^^Just for $hits and giggles I just ran mine:

View attachment 786565

Lou

Have you installed the Mojave beta yet? Seems as though scores are slightly improved running the newest OS. I also believe handheldgames is running 1600mhz RAM down clocked to 1333mhz. That should bump scores slightly as well (not to mention he has 96GB vs you and me only running 24GB). Here is my score running the dual X5690s:

https://browser.geekbench.com/v4/cpu/10004730

First time I ever cracked 3000 with the dual X5690s. Weird, because with my SP setup, I almost always hit 3200 with one X5690:

https://browser.geekbench.com/v4/cpu/8005381

Wonder why the single CPU scores higher in the single core aspect. Kinda frustrating.

 

[flowrider]

^^^^I can not run Mojave just yet. My GPU is an Nvidia GTX 1080. I need to wait for the Nvidia Web Driver before running 10.14. Hopefully the drivers will be out within 24 hours of the OS release. History shows, sometimes even sooner!

Lou

 

[orph]

dual cpu systems have lots of problems, there always slower than one big chip, even AMD threadriper has problems from being kind of one chip kind of not one chip.

threadriper -

data on wrong catch or ram stick ect lots of time spent talking between chips
dual cpu fun
https://www.pugetsystems.com/labs/articles/Should-you-use-a-Dual-Xeon-for-Premiere-Pro-CC-2017-932/

dual cpu setups can give a lot and take a lot

dont stair at benchmarks, try the cpu out, see what it's like.
did you try the factory cpu out? do before the upgrade so you can see how much faster it is (thats half the fun).
if you want after 1 month of use you can still sell it and get a simpler cpu, or just keep it knowing there is no faster cpu for your cmp.

sounds like you have one sweet setup, have fun ^^

and ram, you dont need to upgrade it till you use more than you have, try it out, see how it works with the work you want to do and that will reveal what parts need upgrading or if you ok with it.

 

[kohlson]

dual cpu systems have lots of problems, there always slower than one big chip,

Always...?

NUMA and processor affinity can get pretty complex, especially wrt improving end-to-end performance. AFAIK in MacOS X NUMA/affinity is handled at the OS level. How an application takes advantage of system resources - such as GPU, processor cores, I/O - is determined not only by the developer but also by the OS. And as the linked video discusses, maximizing performance in a multi-processor, multi-core environment is continuing to evolve. And as @orpheus1120 points out, part of the fun in a cMP is trying things out for yourself, and seeing what works!

 

[handheldgames]

Thanks for all the replies!

Honestly, I didn't think a pair of X5677s would score 20000... let alone 21000. Seeing that score confirmed my resolve to purchase a pair and at least test them out. Thanks for that link.

Samsung ecc RDimms delivers exceptional performance. Allowing Geekbench scores in the top 1%, if not higher.

My dual x5680 is running downclocked pc3-12800 Samsung ram. The dual x5677 is running Samsung pc3-10600 ram. Both are at 1333mhz.

dual cpu systems have lots of problems, there always slower than one big chip, even AMD threadriper has problems from being kind of one chip kind of not one chip.

threadriper -

data on wrong catch or ram stick ect lots of time spent talking between chips
dual cpu fun
https://www.pugetsystems.com/labs/articles/Should-you-use-a-Dual-Xeon-for-Premiere-Pro-CC-2017-932/

dual cpu setups can give a lot and take a lot

dont stair at benchmarks, try the cpu out, see what it's like.
did you try the factory cpu out? do before the upgrade so you can see how much faster it is (thats half the fun).
if you want after 1 month of use you can still sell it and get a simpler cpu, or just keep it knowing there is no faster cpu for your cmp.

sounds like you have one sweet setup, have fun ^^

and ram, you dont need to upgrade it till you use more than you have, try it out, see how it works with the work you want to do and that will reveal what parts need upgrading or if you ok with it.

MacOS does a stellar job at taking advantage of additional CPU cores. Windows is a completely different beast and not an applicable comparison to grand central dispatch and asynchronous process on MacOS.

 

[orph]

im not saying it's bad to have dual cpu's im just wanted to point out it's not perfect scaling ^^
there are some real advantages to dual cpu setups

 

[handheldgames]

im not saying it's bad to have dual cpu's im just wanted to point out it's not perfect scaling ^^
there are some real advantages to dual cpu setups

Agreed. Comparing dual to single cpus...

Memory speed suffers in dual cpu synthetic tests, although you get much more addressable ram.
Single core Drops by 7%, but most apps leverage multiple cores.
Multi-core scales by 90%. While not a 2:2 speed gain, 2:1.9 is a great gain. The key here is using great ram. Aka samsung. Not expensive ram.
Drops in single core / imperfect multi core scaling is due to the drop in memory performance from dual cpu overhead.

In a production setting, when you are doing more than 1 task, across multiple workflows, dual cpu’s deliver bandwidth where a single cpu system falls flat.

 

[tpivette89]

Samsung ecc RDimms delivers exceptional performance. Allowing Geekbench scores in the top 1%, if not higher.

My dual x5680 is running downclocked pc3-12800 Samsung ram. The dual x5677 is running Samsung pc3-10600 ram. Both are at 1333mhz.

Didn't know Samsung DIMMs deliver better than average performance. In fact, I just read a few threads where members attested to them giving off more heat than, say, Hynix or Micron. Was leaning more towards Micron for my "wish list future upgrade" 16GB DIMMs due to this, but now I may have to do some more research.

My pair of X5677s just arrived in the mail today. Will delid them and install and start the benchmarking tomorrow. Can't wait to see the results!

 

[tu2thepoo]

I plan to use the computer to run either iMovie or FCPX... would the extra 4 cores for $142 be worth it (non-professional) vs the 8 cores running the same clock speed?

FCPX might be better about this, but when I was making 4k timelapses a couple months ago I noticed iMovie wasn't using all six cores of my W3690. I think it topped out at around 400% CPU use during export (I wasn't checking CPU use while in the timeline view). This was importing/exporting 4k H.264 video and adding basic transitions+background music, and iMovie was using VTDecoderXPCService/VTEncoderXPCService.

Again, that was just my experience using iMovie with that specific workflow. Handbrake, for example, goes up to the full ~1200% CPU use (6 physical cores+6 logical cores).

 

[handheldgames]

Didn't know Samsung DIMMs deliver better than average performance. In fact, I just read a few threads where members attested to them giving off more heat than, say, Hynix or Micron. Was leaning more towards Micron for my "wish list future upgrade" 16GB DIMMs due to this, but now I may have to do some more research.

My pair of X5677s just arrived in the mail today. Will delid them and install and start the benchmarking tomorrow. Can't wait to see the results!

Irrespective of what the pundits say, the 48gb and 96gb combinations of Samsung ram I’m using hits the top 1%, if not the top Geekbench score for an unoverclocked Mac Pro in MacOS in single and dual cpus.

 

[orph]

im not shore what parts you have or dont have at the mo but just try editing video, it may be as fast as you need. dont get to obsessed in "i need part X to get 10% more potential speed" etc that time is better spent learning
learn to plan a shoot
learn how to shoot it
learn how to edit it (both the skill of editing and the software)
learn how to publish and promote (and promote fast that can take so much time)

learn to see you mistakes and improve

read up on contracts, copyright and laws
where can i shot video
who can i shoot
etc

and never forget audio is relay important, not god level audio but good audio !

try davinci resolve (it has a free version that is relay amazing)
https://www.blackmagicdesign.com/uk/products/davinciresolve/

o and start keeping your software and camera manuals pdf files on your phone

edit
that got long, what i wanted to say is get good with what you have then once you relay need to get the super fancy computer parts i know pro's still on basic setups, they do work not computers

 

[XNorth]

I upgraded to 12 cores X5690s for FCPX. It's a worthy investment considering I will continue using the cMP for as long as possible.

BruceX test: 16-17 seconds

Geekbench scores....

 

[tpivette89]

My pair of X5677s just arrived in the mail today. Will delid them and install and start the benchmarking tomorrow. Can't wait to see the results!

Damn... I hate to admit this, but the de-lidding process didn't go as well as the dozen or so of the other ones I have done. I used a different vice than I normally do, and things took a turn for the worst. One of the X5677s crushed a bit on the edge, and the other one bent before the lid popped off. I've NEVER had an issue doing this procedure, and for now, I have to attribute it to an unfamiliar tool I used.

I tried to install the CPUs in the Mac Pro anyway, and it did not boot. Re-installed the pair of X5690s I have been running before the experiment, and it did boot, with Geekbench scores intact:

https://browser.geekbench.com/v4/cpu/10044100

Wish I could have ran the pair of X5677s... due to their cheap pricing, I may try another pair and de-lid them on the vice I typically have done it. But for now, I guess I'm sticking with the 12-core setup.